Author: Dann Corbit
Date: 09:58:59 09/11/01
Go up one level in this thread
On September 11, 2001 at 08:45:09, Andrew Williams wrote: [snip] >Can you confirm this? I had both binaries. >Your friend's program is *nearly twice as fast* under >MSVC as under cygwin? I don't know for sure if it was Cygwin, but I have seen speedups like that before. > This is with proper optimization in both cases, I >presume? I don't know exactly how the original was built. That was just one instance. >I ask because I don't use Windows at home, and have always considered it would >be a complete waste of money, considering I'd have to buy Windows and the >compiler. But having my program run twice as fast might begin to be tempting. There is a good alternative for Linux. I believe that KAP has a compiler for Linux now. That should be worth a try. And I think Intel may be doing a port to Linux of their compiler also. It is a fact that GCC is not as good as commercial compilers. It does have the best error diagnostics of any compiler I have used, however. I sometimes use it as a sort of lint. I had high hopes for GCC 3.0 but that was a real bust. I did not see any of the promised performance boost -- same sort of output as 2.95. Some UNIX programs are difficult to port. If you use fork(), alarm() and POSIX stuff like that you will have lots of trouble trying to make a good PC version. The emulators like Cygwin are incomplete, and some things (such as fork()) are horribly slow. On the other hand, if the code is mostly ANSI/ISO C, the port can usually be made in a very short period of time. Instead of going out and buying a big pile of stuff, go to a friend's house who has all the equipment and compilers and try it out first. Then, if you get a huge boost, you might like to switch. Of course, it is the algorithm that dominates anyway. A doubling in speed is only 50 ELO or so (but every little bit helps).
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.